The DiPOLE project: towards high energy, high repetition rate diode pumped

Contact [email protected]

Klaus Ertel, Saumyabrata Banerjee, Paul Mason, Jonathan Tristan Davenne, Michael Fitton, John Hill, Andrew Phillips, Phil Rice, Steph Tomlinson, Christian Sawyer, Lintern Steve Blake, Cristina Hernandez-Gomez, John Collier Engineering Technology Centre, STFC Rutherford Appleton Central Facility, STFC Rutherford Appleton Laboratory, Laboratory, Chilton, Didcot, OX11 0QX, UK Chilton, Didcot, OX11 0QX, UK

Introduction material selection DiPOLE stands for Di ode Pumped Optical Laser for The starting point was the identification of the most suitable Experiments. It is a new project at the CLF to develop the gain material. This material needs to provide: foundations of novel high energy, high average power laser • A long lifetime to minimise the number of systems based on diode pumped solid state laser (DPSSL) pump diodes required technology. Compared to conventional systems, this approach promises dramatically increased repetition rates (and hence • Good thermo-mechanical properties to handle the high average powers) at significantly higher electrical-to-optical average power conversion efficiency. DiPOLE has been included as an • Reasonably high gain cross section to enable uncomplicated emerging opportunity in the Research Councils UK Large and efficient energy extraction Facilities Roadmap [1]. • The possibility to be manufactured in large sizes to handle Motivation the high pulse energy. Laser capable of producing energetic ns-pulses are (Yb) as an active laser offers very long one of the main tools for laser plasma research and high-energy fluorescence lifetimes, a low quantum defect (pump wavelength applications. Laser chains containing such amplifiers can 940 nm, laser wavelength 1030 nm), reasonable gain cross produce ns-pulses or ps-pulses if the chirped pulse amplification sections and efficient high power laser diodes are readily (CPA) technique is used. Depending on the application, these available for its pump wavelength. Two host materials have pulses are either applied directly or are used to pump other been identified that offer good thermo-mechanical properties amplifiers (e.g. Ti: or OPCPA) in order to obtain even and can be manufactured in large sizes: crystalline calcium shorter pulses in the fs-regime. fluoride (CAF) and ceramic YAG. Currently, ns-amplifiers are based on flashlamp-pumped Yb:CAF has a very long fluorescence lifetime of 2.4 ms and a Nd: technology and their repetition rate is limited to a few large gain bandwidth (> 50 nm) [4], and is therefore a shots per minute for amplifiers delivering tens of joules of pulse promising candidate for directly diode pumped chirped pulse energy to a few shots per day for lasers delivering kJ-level pulse amplification (CPA) systems for producing sub-ps pulses [5]. energies. However, since it exhibits a very small gain cross section, very Increasing the repetition rate of such laser systems to the multi- large fluence levels are required for both pumping and Hz level (typically 10 Hz) is pivotal for the following extraction in order to achieve good optical-to-optical (o-o) applications: efficiency. • Opening up new horizons in fundamental laser plasma interaction research by enabling higher throughput and the exploration of larger parameter spaces. • So-called secondary sources which use laser-generated plasmas to produce ultra-short pulses of energetic particles (electrons or ) or electromagnetic radiation (ranging from THz to hard X-ray). High repetition rate drive lasers are required to generate sufficiently high particle and photon numbers. Much of the pan-European ELI project focuses on the development and exploitation of secondary sources [2]. • Inertial confinement fusion (ICF), which is expected to be demonstrated for the first time within the next two years. Whereas current low-repetition rate facilities like NIF and Fig. 1: Yb:YAG - Cr 4+ :YAG compound disk. LMJ are suitable for proof of principle experiments, high- efficiency, high repetition rate DPSSL based laser drivers On the other hand, Yb:YAG has an order of magnitude higher open up the possibility to develop ICF into a reasonably gain cross section with a reasonable fluorescence lifetime of clean, practically inexhaustible source of energy. This is the 1 ms [6]. Since the main application of our envisioned kJ-class focus of the pan-European HiPER project [3]. laser is the production of ns-pulses, either for pumping amplifiers for fs-pulse generation (Ti:sapphire or OPCPA) or concept for driving inertial fusion targets, we think that ceramic The main activity within DiPOLE is the development of a Yb:YAG is the best choice for the gain medium. Also, DPSSL amplifier concept that is capable of delivering kJ-level monolithic compound structures with different doping species pulses at 10 Hz repetition rate. are possible with ceramic YAG [7]. The photo of such a compound disk is shown in Fig. 1. Here the inner region of the therefore chosen as the preliminary operating point for our disk is doped with Yb 3+ and acts as the active laser medium, the amplifier. outer region is doped with Cr 4+ which heavily absorbs at 1030 nm and therefore acts as an index-matched absorber for suppression of amplified spontaneous emission (ASE) [8]. Efficiency and gain modelling Numerical modelling has been carried out to determine optimum amplifier design parameters. In the model, the storage efficiency ηstor has been calculated for various parameters like pump fluence, pump pulse duration and pump spectral width. ηstor is defined as extractable fluence divided by pump fluence. Loss mechanisms and associated efficiency factors that influence ηstor are: ηfluo for the fluorescence decay, ηQD for the quantum defect, ηabs for pump light that passes through the gain medium without being absorbed and, finally, ηreabs for the minimum upper state population that needs to be established in order to overcome reabsorption that exists due to the quasi-3- Fig. 3: Maximum storage efficiency (blue) and small signal level nature of Yb:YAG. If a pump pulse duration of 1 ms is gain (red) for amplifier operated at 175 K. chosen, η and η limit η to 58 %. It turns out that η QD fluo stor abs Operating at cryogenic temperatures yields the added benefit of and η need to be balanced off against each other and that for reabs improved thermo-mechanical and thermo-optical properties like a given set of pump-related parameters, there is one optimum increased thermal conductivity and reduced temperature value for gain medium optical depth (OD = thickness times dependence of the [10]. Another effect is the doping concentration) that yields the maximum η . If OD is stor narrowing of spectral features both in the emission and the chosen too low, too little pump light is absorbed, if OD is too absorption spectrum. The absorption spectra measured at room high, reabsorption losses become dominant. temperature and at 175 K are shown Fig. 4, together with a The following results are calculated, unless stated otherwise, for 5 nm FWHM Gaussian spectrum for comparison, which is an amplifier that is end-pumped from both sides with a pump assumed to be the spectral shape of our pump source. pulse duration of 1 ms, a 5 nm FWHM pump spectral width, centred at the optimum wavelength in the 940 nm absorption band. Spectrally resolved pump absorption cross sections were taken from [9]. Quantities calculated were ηstor and the small signal gain G, defined as G = exp( ηstor Fpump /F sat ) where ηstor Fpump is the extractable fluence and F sat the gain saturation fluence. First, calculations were carried out for room temperature operation. The results are shown Fig. 2. It becomes apparent that very strong pumping is required, firstly to overcome the high reabsorption losses and to achieve good efficiency and secondly to overcome the still rather low gain cross section and achieve reasonable gain. The required high pump and extraction fluences are difficult to achieve because of limited pump source brightness and limited . Fig. 4: Absorption spectra of Yb:YAG at different temperatures and 5 nm wide pump diode spectrum for comparison.

Fig. 2: Maximum storage efficiency (blue) and small signal gain (red) for amplifier operated at room temperature.

Cooling the gain medium to 175 K drastically changes the Fig. 5: Storage efficiency as function of pump centre wave- situation, as illustrated in Fig. 3. Reabsorption is reduced and length for two different temperatures and pump fluences. the gain cross section increased, leading to greatly improved efficiency and gain, especially at moderate fluences. A pump If these pump and absorption spectra are used to calculate fluence that is realistically achievable with today’s laser diodes storage efficiency for two different temperature scenarios, is 10 J/cm 2 (5 J/cm 2 from each side), yielding a storage results as shown in Fig. 5 are obtained. Even though the pump efficiency of just over 50 % (resulting in an extractable fluence fluence in the low temperature scenario is only half that of the of 5 J/cm 2) and a small signal gain of 3.8. This fluence is room temperature case, significantly higher storage efficiency is predicted. Lower temperature operation also shows a much weaker dependence on pump centre wavelength. So despite DiPOLE prototype narrower absorption features, the requirements with respect to To test the concept in the laboratory a lower-energy multi-J spectral performance of the pump diodes are less critical for low prototype amplifier system is currently being built. The design temperature operation. is based on four co-sintered ceramic YAG discs (55 mm in Amplifier geometry diameter x 5 mm thick) where the Yb-doped region (35 mm diameter) is surrounded by a 10 mm thick Cr 4+ cladding to After determining operating temperature and pump fluence for absorb unwanted transverse fluorescence. A photograph of one our envisioned amplifier, the actual geometry needs to be of these discs is shown in Fig. 1. Given the 2 cm total gain defined. If the laser system is to yield an output energy of 1 kJ medium thickness, two different Yb doping levels of 1.1 and and the amplifier is to be operated at an output fluence of 2.0 atomic% have been chosen to maximise storage efficiency 5 J/cm 2, the aperture needs to be 200 cm 2 or 14 x 14 cm 2 if a and to equalise the heat loading. The discs are mounted in a square beam shape is adopted. The optimum OD = Nxl obtained vacuum insulated pressure vessel through which cryogenically from the numerical calculations is 3.15 % cm, where N is the cooled He is flowed. A schematic of the cryocooler system Yb-doping concentration in atomic % and l the geometrical is shown in Fig. 8. The amplifier is end-pumped from both sides thickness of the amplifier. The choice of N and consequently l is by two diode pump lasers operating near 940 nm. Each source governed by ASE management considerations. If the gain- provides 20 kW peak power in pulses of ~1 ms duration in a length product along the diagonal across the (square) surface of square beam (2 x 2 cm2), with a corresponding pump intensity the amplifier is to be kept below 3, we require N < 0.18 % and of 5kW/cm 2 and at a repetition rate of 10 Hz. hence l > 18 cm. Such a thick amplifier requires distributed cooling as demonstrated on the [11]. There, the gain medium is divided into a stack of thin slabs with He gas flowing through the gaps between the slabs. The concept is illustrated in Fig. 6, where an amplifier consisting of 10 slabs is shown.

Fig. 8: Schematic diagram of the cryocooler system under development at CLF.

Fig. 6: Illustration of amplifier geometry: isometric (left) and Conclusions side view (right). In summary, we have presented the conceptual design of a If the criterion that the transverse gain-length product must not cryogenic Yb:YAG amplifier that can be scaled to kJ energy levels and beyond, owing to its geometry and unique cooling exceed a certain value is applied to each individual slab, one realises that the doping concentration can be increased towards technique. Considerable enhancement in optical-to-optical the centre of the amplifier. The advantage is twofold: firstly, conversion efficiency is predicted with the reduction of pump since the required overall OD remains the same, the amplifier as fluence at low temperatures owing to the reduced reabsorption a whole becomes thinner, saving material and reducing the losses, increased pump adsorption and gain cross sections. impact of nonlinear effects, and secondly the optical power Numerical modelling also shows a weaker dependence on pump absorbed in the individual slabs and hence the heat load is centre wavelength at low temperature, relaxing the stringent equalised. An optimised doping profile for a 10-slab amplifier is requirements of the diode pump source. The introduction of shown in Fig 7, together with the transverse gain as a function doping gradient in the slab architecture helps facilitate a of position. The overall thickness of the gain medium is reduced uniform gain distribution and heat load in addition to the from 18 cm to 10 cm in this configuration. reduction of the overall thickness of the amplifier medium from 18 cm to 10 cm. A lower-energy multi-J prototype amplifier system is currently under development at CLF as a proof of principle.

References 1. Research Councils UK Large Facilities Roadmap, http://www.rcuk.ac.uk/research/resinfra/lfroadmap.htm . 2. The Extreme Light Infrastructure (ELI) European Project, http://www.extreme-light-infrastructure.eu/ . 3. HiPER project, http://www.hiper-laser.org/ . 4. P. Camy et al, "Comparative spectroscopic and laser 3+ properties of Yb -doped CaF 2, SrF 2 and BaF 2 single Fig. 7: Transverse gain-length product and doping levels along crystals," Appl. Phys. B 89 , 539–542 (2007). optical axis in 10-slab amplifier. 5. M. Siebold et al, "Terawatt diode-pumped Yb:CaF 2 laser," Opt. Lett. 33, 2770-2772 (2008). 6. T. Taira, "RE 3+ -Ion-Doped YAG Ceramic Lasers," IEEE J. Sel. Top. Quant. 13 , 798-809 (2007).

7. H. Yagi, " Y3Al5 O12 ceramic absorbers for the suppression of parasitic oscillation in high-power Nd:YAG lasers, " J. Luminescence 121 , 88-94 (2006). 8. K. Ertel et al, "ASE suppression in a high energy sapphire amplifier," Opt. Express 16 , 8039-8049 (2008). 9. D. C. Brown et al, "Yb:YAG absorption at ambient and cryogenic temperatures, " IEEE J. Sel. Top. Quant. 11, 604- 612 (2007). 10. T. Y. Fan et al, "Cryogenic Yb 3+ -doped solid-state lasers," IEEE J. Sel. Top. Quant. 13, 448-459 (2007). 11. A. Bayramian et al, "The mercury project: A high average power, gas-cooled laser for inertial fusion energy development," Fusion Sci. Technol. 52 , 383-387 (2007). Timing and Synchronisation System Designs for the New Light Source

Contact [email protected]

G. J. Hirst i STFC Central Laser Facility Rutherford Appleton Laboratory, HSIC, Didcot OX11 0QX, UK

Introduction ×3 The New Light Source (NLS) [1] was a proposed UK and international user facility based around a combination of Free 650 MHz phase PLL 216.67 MHz Electron Lasers (FELs) and conventional lasers. In 2009 it was fibre laser sensor electronics decided to halt work on the proposal once the Conceptual oscillator Design Report had been completed. Among the subsystems described in the CDR [1] was a sophisticated laser-based timing 8 and synchronisation network, designed by CLF staff. ÷ 2 0.846 MHz The NLS was designed to support a very wide range of science, ÷ 2 3 105.8 kHz a significant fraction of which exploited its ultrashort pulse capabilities and, in particular, its ability to deliver several 10 MHz beams to an experiment simultaneously. The relative timing of × 2 6 RF reference ÷ (2 2×3) 8.816 kHz oscillator the pulses in these beams had to be tightly managed, either to overlap them reliably in time, e.g. for multiphoton experiments, ÷ 2 3 1.102 kHz or to provide them with variable relative delay(s) e.g. for pump-probe studies. In addition to the lasers delivering beams Figure 1. NLS master clock architecture showing pulse rate directly to users there were also lasers acting as components of division scheme. the NLS machine itself. The timing of the pulses from these needed to be controlled to ensure that the facility functioned at Synchronisation all. This article describes the systems used to manage the timing The potential that ultrashort pulses offer for very high temporal and considers the performance of an example subsystem in resolution and also for high-intensity multiphoton science can more detail. only be achieved if the relative pulse timing is very tightly Time Structure managed. For a subset of NLS experiments it would be sufficient if the timing management consisted simply of The NLS time structure was decided in consultation with the pulse-by-pulse timing measurement followed by time-stamping scientific user community. The outcome was particularly of the resulting data. However there were also experiments for simple, driven by a strong preference for the pulses from each which this approach would not be adequate and, in particular, FEL and laser to be delivered in a continuous, evenly-spaced, there were aspects of the machine operation (for example the high-rate train. (Experiments needing a lower rate would have use of conventional lasers to seed the FELs) where the timing the option of pulse selection.) As far as the timing was would need to be controlled rather than just measured. For this concerned the only parameters to be decided were, therefore, reason a synchronisation control system was designed which the pulse duration and repetition rate. The duration was chosen would be deployed across the whole of the NLS. to be 20 fs FWHM to allow the best possible time resolution. There was a clear science requirement for an initial rate of at Given the 20 fs FWHM chosen pulse length, the target figure least 1 kHz with an upgrade path that would increase this, in for overall synchronisation was set at 10 fs RMS. This was approximately decade steps, to 1 MHz. In principle exactly recognized as a challenge since individual subsystems have these rates could have been chosen. But in fact slightly different only been developed to this level of performance in recent rates were selected with a view to minimising the timing jitter years. However the NLS would have an advantage over almost of the resulting pulses. all similar facilities in that it would be built from scratch without having to accommodate any pre-existing hardware or The starting point for the rate was the frequency of the RF design decisions. This would allow the need for low timing modules which accelerated the electrons for the FELs. This was jitter to be factored into the design from the start. predetermined by industry standards to be 1.300 GHz. The NLS operating rate would need to be an integer sub-multiple of this The entire synchronisation system is too complex to describe and for technical reasons the timing jitter would be minimised if here. Instead the general approach will be illustrated by an the integer chosen had only very small integer factors itself example. One of the features of the NLS was the configuration (ideally only 2 and 3). The eventual scheme is shown in Fig. 1. of its FELs as seeded amplifiers with the seed pulses derived by The overall master clock frequency was chosen to be high from conventional lasers. The timing 216.67 MHz, the sixth harmonic of which would generate the of the FEL pulses would then be set by the timing of the seed 1.3 GHz RF. The nominal 1 kHz pulse rate to be delivered on lasers and the stability of the output beam transport paths. This Day 1 of operations was actually 1.102 kHz (1.3 GHz divided scheme imposes two separate synchronisation requirements. by 2 17 × 32). Subsequent upgrades would increase it by factors Firstly that the seed pulses must overlap the electron bunches in of 2 3, 2 2 × 3 and 2 3, eventually reaching 0.846 MHz. An the FEL undulator. The nature of electron accelerators, whose transport systems are inherently dispersive, makes this difficult additional feature of this choice was that the rate after each [2] upgrade would be an integer multiple of the rate before. This to achieve. The fact that the NLS accelerator did achieve it is would make it easier for legacy systems to remain operable testament to the very considerable effort and creativity of the after any upgrade. design team. But the bulk of that work concerned the accelerator itself rather than the synchronisation system and it will therefore not be discussed further here.

Figure 2. Layout of the elements used to synchronise FEL and laser pulses at the users’ experiments. 1 and 2 represent the first and second phase sensors for the FEL seed laser. The second challenge is the delivery of the FEL pulses to the Once the seed laser is phase-locked FEL action can begin and users’ experiments in exact synchronism with pulses from a the opportunity then exists to correct timing jitter arising in the conventional “endstation” laser. The scheme devised to do this FEL and beam transport paths. This will involve sensing the is shown in Fig. 2. The physical scale of the system is large - pulse timing after the FEL, at point 2 in Fig. 2. This will be the seed laser and the endstation laser are more than 100 m considerably more challenging than sensing the laser timing apart. Yet the need for less than 10 fs timing jitter corresponds a) because phase sensing in the EUV, where the FEL operates, to a relative movement of the pulses in space of less than 3 µm. is not yet a mature technology, b) because the FEL will be So issues around the transport of beams and of timing signals tuned over a much wider photon energy range than the laser and are critical. For this reason a master clock architecture was c) because the uncorrected noise levels (both amplitude and chosen (the details of the low-noise clock are shown in Fig.1). phase) from the FEL will be higher than for the laser systems The alternative would have been to use a common laser alone. The effectiveness of the correction system depends oscillator and to split the beam from this to feed the two directly on these levels. However despite these problems a amplifier chains. The first problem with this is that transporting post-FEL correction scheme has been included in the design as ultrashort pulses over long distances is difficult. To avoid it promises the easiest route to reaching the overall jitter target. nonlinear effects the transport has to be in free-space (ideally in The tuning issue will also apply to the endstation laser if it vacuum) using curved mirrors for image-relaying to preserve proves necessary to stabilise its timing after its nonlinear the beam quality. Engineering such a system for high stability frequency conversion. The initial plan is to achieve stability would be very expensive. On the other hand the technology passively, by mounting the converters close to the endstation needed to transport optical clock pulses with few-femtosecond and in a well-controlled environment. But the option of active jitter has already been developed and demonstrated[3]. stabilisation has not been discounted.

The second problem with a common oscillator is that the active Conclusions and Acknowledgement control system needed to correct the pulse timing in real time Some elements of the timing and synchronisation scheme for ideally uses the position of an oscillator cavity mirror as its the NLS have been described. The timing is expected to be actuator. Such a mirror is multi-passed very many times on the straightforward. Delivering 10 fs RMS overall synchronism will timescale of the correction signal (typically >10 µs). So the be challenging, but it should be achievable using state of the art mirror need only be moved a tiny distance, which eases high subsystems. This work was partly supported by “IRUVX-PP”, speed operation. Of course if there was only one oscillator then an EU co-funded project under FP7 (Grant Agreement 211285). movement of its mirrors would not correct the relative timing of the endstation laser and FEL pulses. So an extra-cavity actuator References would have to be used which would be less effective. 1. J. Marangos et al, NLS Project: Conceptual Design Report , Having distributed accurate clock signals to the phase (timing) publ. STFC (2010) sensors in Fig. 2, the next task is to reduce the timing error they detect using a closed phase-locked loop (PLL). Again this is a 2. R. Bartolini, Optimisation of a single-pass superconducting well-established technique and locking of a laser oscillator to an linac as a FEL driver for the NLS project , Paper WEOB02, st external reference using optical phase sensing has also been 31 International FEL Conference, Liverpool, UK (2009) [4] demonstrated at the femtosecond level . Extending this 3. J. Kim et al, Long-term femtosecond timing link technique to include the laser amplifier chain inside the timing stabilization using a single-crystal balanced cross control loop should be straightforward, the only complication correlator , Opt Letts 32 (9) 1044 (2007) being the reduction in pulse rate relative to the oscillator. This imposes a Nyquist limit on the phase noise spectrum which can 4. See e.g. T. R. Schibli et al, Attosecond active be sensed. However given that the lasers are designed to operate synchronization of passively mode-locked lasers by at up to 1 MHz it seems likely that this limit will be higher than balanced cross-correlation , Opt Letts 28 (11) 947 (2003) the one imposed by the correction actuators. Experimental setup in the Vulcan HaPPIE Laboratory for Multi-beam Combination to achieve Diffraction limit pulses

Contact [email protected]

P.J. Phillips, C. Hernandez-Gomez, J. Collier CLF, Science and Technology Facilities Council Rutherford Appleton Laboratory Harwell Science and Innovation Campus Didcot OX11 0QX

Introduction measure the wavefront of one aperture and correct this High powered lasers are attractive owing to their potential for a wavefront. The two most common methods is a Shack- Hartmann wavefront sensor and an interferometer based diverse range of experiments. In order to achieve these high 4 powered lasers, various choices for amplification of laser beam technique via the multiwave lateral shearing interferometer . A have to be made. These choices include the size of the beam Shack-Hartmann technique samples the wavefront with the use which in turn requires laser crystals or ceramics to be of a grid of microlenses which is focused onto a sensor, manufactured close to their current limit of scalability. Another generally a CCD camera. In the case of a distorted wavefront way of achieving these high power requirements is to combine the focused points on the CCD camera are misaligned relative several beams into a monolithic beam, which immediately to an undistorted wavefront which is proportional to the reduces the requirements for the amplifier to a more modest distortion in the wavefront. The position of focused spots level. This requires technology to lock the beams spatially and compared to a set of reference focused points for a “perfect” temporally. wavefront is used create a phase map can be extracted of the measured wavefront or a Fourier demodulation technique, A number of projects are underway in Europe under the which obtains an error signal. The phase map for one aperture is umbrella of the LASERLAB-EUROPE II to research high generally expressed as a set of either Zernike or Legendre power laser beams. There are several initiatives under this polynomials. Error signal can then be used to correct the known as Joint Research Areas (JRA) of which one of them is wavefront by the use of an adaptive mirror. Adaptive mirror the High Average and Peak Power lasers for Interaction contains actuators to control the shape of the mirror so that the Experiments (HAPPIE) to target the technology areas which are distortion can be corrected in the wavefront. In the case a deemed useful to achieve high power lasers. One of these multiwave lateral shearing interferometer uses a 2D diffraction technology areas is the control of the spatial and temporal grating which replicates the incident beam into 4 identical overlap of combining these beams in a coherent manner. waves which are propagated along slightly different directions. Specifically in the case of laser fusion project, known as These directions differences create inference patterns. After a HiPER 1 it is envisaged to include a fast ignition beam. This fast few millimeters propagation the four beams are slightly pulse ignition beam requires ~60 KJ with Picoseconds pulse separated. The grid deformations are directly connected to the width. This beam is to be composed of multiple subbeams in phase gradients. A spectral analysis using Fourier transforms order to achieve the required power levels for fast ignition. A allows the phase gradient extraction in two orthogonal planes. number of these subbeams will be combined to act as The phase map is finally obtained by integration of these monolithic beam. This also reduces the requirements on the gradients. As explained in the case of Shack-Hartmann sensor power from each individual amplifier stage and gives a level of the phase map can be used as an error signal to an adaptive control on the different subbeams for pulse shaping. In order for mirror. These methods do not provide a piston error, this being the laser power to achieve the diffraction limit at the interaction the amount of lateral movement one beam has to make with the point, it requires that the subbeams be temporally and spatial second beam so that they are spatial overlapped as they overlapped. It is also currently thought that the main igniter propagate, so cannot be used to spatial overlap two beams. beam would now also compose of subbeams. Another pan European laser project called Extreme Light To create a phase difference between two independent sub Interaction (ELI); which involves development and construction apertures then it is important to interfere the two wavefronts. of three separate lasers in different geographical locations. It is This then creates an interference pattern between the two currently thought that each laser will be constructed from wavefronts from the sub apertures and then the phase several laser beams which also require spatial and temporal information (piston measurement) can be extracted. In order to obtain this piston measurement between two sub aperture beams overlap. 4 2 then a quadriwave lateral shearing interferometer (QWLSI) Spatial overlapping can be achieved by several techniques ; can be employed. This measurement is self referenced, which most prominently this is achieved in the Large Telescopes makes it easy to implement. This method adapts the well known either in use or currently being developed around the world. multiwave lateral shearing interferometry used for wavefront These large telescopes are constructed from segmented mirrors sensing to phase difference evaluation by extending the and the wavefront needs to be flattened to within a fraction of shearing distance so that the edge of one subbeam overlaps the the wavelength to achieve the diffraction limit to fully exploit edge of another subbeam. Analysing the interference pattern in the capability of the telescope. We are currently setting up a this overlap region leads to this phase difference between the laboratory to investigate the different techniques in wavefront two subbeams. One particular system designed to do this is the measurement spatial and temporally. PHASICS camera SID4. This was developed in collaboration with the CEA CESTA who developed the system for the 5 Wavefront measurement methodologies phasing of two LIL compressed beams . The piston accuracy can be estimated to be 40 nm or less. In the case of wavefront measurement there are several different methods which are employed. These methods traditional If we consider two beams with complex amplitude AʚBʛ Ɣ ψ ǭI͵ʚͶʛexpʚjʚ ͵ʚͶʛʛ ƍ c. c. then split the beam into a 4 wave interferometry. We get a case of independent subbeams, and to obtain the simplest beamlet combination, it is ne cessary to align the physical gap direction between the subbeams with the interferometer replication grating in axis y. In this case we obtain a situation similar to the 2-wave case. We observe 4 - wave interferences, with the two wave interference zones at its left and right. The 4-wave interference zone labelled OR is illuminated by four replicas, A1, A3, B2 and B4. In the OR zone , the pattern results from the amplitude double ∗ ∗ product ʚAͥBͦ ƍ AͧBͨʛ. The argument of this sum leads to a phase difference that is equal to ʞʚψ ƍ ψ ʛ Ǝ ʚψ ƍ ͵ͥ ͵ͧ Ͷͦ ψͶͨʛʟ/2 Ɩ ψ͵ Ǝ ψͶ. In the left (right) zone, the pattern results Figure 1. Experimental layout. from the interference of replicas from the same initial beam A (B). The argument of the corresponding amplitude sum leads to A (B) beam phase gradient. Considering an un aberrated beam, this phase gradient should be equal to zero or to a constant if the wavefront is tilted. In the case of unaberrated beams a graph of the total amplitude argument versus x may represent a top hat shape where the amplitude, P(x), represents t he phase difference between the two independent subbeams A and B. Futhermore, in the case of angularly phased subbeams, its amplitude would be equal to the piston difference between them. A deeper theoretical study shows that a deconvoluted signal in the overlap zone allows characterization of not only the piston but also tip and tilt and higher order aberrations.

Current Progress Figure 2. Laboratory setup, beam path is shown by the use of the red line. Under the JRA there is a specific task that requires the development of novel metrology and techniques to lock a The current PHASICS software on the laptop displays the two number of pul ses of discreet test large apertures together subbeam wavefront measurement plus interference between the through active phase control. These test apertures are of a small two sub beams. The interference area can adjusted by varying size so that any technical difficulties can be solved at this point distance between the grating and the CCD sensor up to a in so that they can be scaled to a larger size required for current maximum. The distance is defined by the beam parameters and planned facilities. With this in mind at CLF we have developed the distance that separates the two beams. An algorithm is then a laboratory to asses various techniques for beam phasing employed to calculate the difference in phase between the two experiments. The initial experiments will conducted in CW beams, which is displayed as a fraction of the wavelength being mode of a Ti:Sapphire laser output. A PHASICS camera is measured. This is carried out by the use of Legendre employed for capture and anal ysis of the combined wavefront. polynomials as opposed to Zernike polynom ials due to the fact As shown in Figure 1 the optical layout expands the beam to a that the wavefronts are square. The wavefronts can be 120 mm beam from the laser , so that it can be subdivided into at measured with the use of Zernike polynomials by defining the maximum four beams of 50mm square. Although initially the area as circular in the software. The piston and discrete Tip / experiment will start by using two beam s of 50 mm square Tilt errors does not yet go to the translation stage an d their each. The two mirrors have independent control of tip / tilt and respective stages. With the purchase of a programming a translation stage for each mirror for piston control. Mirror 1 language from PHASCIS, which will gives us access to the is the slave to Mirror 2 so that when a piston error is calculated analyzing software and then it will be possible to introduce a by the PHASICS camera and software, this translation stage is feedback loop to control the piston error and the discrete Tip / aligned to bring the piston error close to zero. I n Figure 1 the Tilt, which is what we are planning to do in the immediate combined beam is relay imaged from the two square mirrors future. onto the PHASICS camera, to ensure that the plane of reference This consists of a Ti:Sapphire Tsunami working from 0.8 to 1.1 is imaged onto the focal plane array of the came ra. The beam µm, 100 fs pulses at 80 MHz pulse rep rate. condenser system composes of a 570 mm lens and a 100 mm lens to collimate the beam. This makes the beam slightly larger Future Developments than the aperture (4.8 mm by 3.8 mm) of the camera. Wavelength range of the sensor is 350 - 1100 nm with a spatial This project is funded by the LASERLAB -EUROPE II fund. resolution of 29.6 µm. There is also a beamsplitter so that the far field im age of the beam so that the diffraction limit of the We will develop the system for analyzing the wavefront for a spot is measured. The PHASICS camera is connected to a feedback control system, so that the two beams will be locked laptop to analyze the wavefronts and in the future implement spatial. This will initially be carried out on a two beam system, the phase correction. Figure 2 actu al shows the arrangement in which will be expanded to a four beam system. This will require the laboratory to date. fur ther mirrors and a different configuration to arrange the mirrors. We also require further development of the software so that three of the mirrors receive error signals to spatial overlap them.

We would like to implement a basic interferometer system so that an estimation of the distance between n numbers of mirrors can be calculated in the first instance. In which the PHASICS camera can then finely adjust the n numbers of mirrors in the system.

Once the system is in a working state and fully tested in the HAPPIE laboratory then we would like to develop the system on the Astra-Gemini laser system here at CLF, to spatial overlap the two beams.

We also like to explore techniques to temporal lock several beams together. Conclusions We have a setup in the HAPPIE laboratory with the current technology to carry out analyses of subbeams wavefront to spatial phase them. In the future we are going to design a feedback system so that the two beams are permanently spatial locked after this we would implement the system on Astra- Gemini. We are also going to explore experimentally the temporal locking two picoseconds pulses in the future.

References 1. A high power laser fusion facility for Europe” Nature Physics, Vol 2, Jan. 2006 2. “Shack-Hartmann sensor for the active phasing experiment” Proc. SPIE 6267, 626730,(2006) 3. “Theoretical description of Shack-Hartmann wavefront sensor” Optics Communication 222 (2003) 81-92 4. “Single-shot wavefront measurements of high intensity ultrashort laser pulses with a three wave inteferoemeter” Optics Letters Vol. 23 No.8 1998 5. “Overview of PETAL, multi-Petawatt project on the LIL facility”, Plasma Physics and controlled Fusion, 50, (2008)124045

6. “Piston measurement by quadriwave lateral shearing interferometry” Optics Letters, Vol.31, No. 17 September 1, 2006